Enabling Scalable Photonic Tensor Cores with Polarization-Domain Photonic Computing

2025-01-31

AI Summary

Conventional coherent optical networks impose fundamental limitations on computational precision in photonic integrated circuits for optical computing. Method: This work proposes a wavelength–polarization dual-degree-of-freedom photonic tensor core architecture, implemented on a silicon photonics platform integrated with a two-dimensional ferroelectric heterostructure (h-BN/α-In₂Se₃). It pioneers the synergistic use of polarization-multiplexed modulation and wavelength-selective routing for tensor operations, coupled with experimentally calibrated device modeling to overcome coherence-detection precision bottlenecks. Contribution/Results: The architecture enables high-throughput, energy-efficient matrix computation on a single chip. Experimental evaluation demonstrates an 83% improvement in computational accuracy over state-of-the-art coherent optical networks, alongside significantly enhanced energy efficiency and scalability—establishing a new paradigm for large-scale photonic neural networks.

Technology Category

• Machine Learning: Matrix & Tensor Methods
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Learning to Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Sustainability and carbon-aware systems for Web, mobile, and WoT
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

We present a silicon-photonic tensor core using 2D ferroelectric materials to enable wavelength- and polarization-domain computing. Results, based on experimentally characterized material properties, show up to 83% improvement in computation accuracy compared to coherent networks.

Problem

• Optical Computing
• Accuracy Improvement
• Color and Direction Sensitivity

Innovation

• Optical Computing Chip
• Color and Direction-based Computation
• Enhanced Accuracy